Quantifying the impact of protocol design on enrollment and dropout rates in breast cancer clinical trials.

11027 Background: Clinical trial enrollment can sometimes fail due to protocol design being overly complex, including procedures that are overly burdensome on patients and/or trial sites. Procedures that may be more logistically challenging or introduce higher risk (ex: radiation exposure in imaging procedures) can affect trial enrollment and patient retention. To better understand factors that optimize operational design of protocols, we evaluated breast cancer protocols to assess and quantify procedures that had the highest potential impact on enrollment and dropout rates. Objective: Our objective is to develop an algorithm that can predict enrollment and dropout rates by analyzing the procedures and procedure frequency included in protocol design. Methods: 42 breast cancer protocols (sourced from aggregated, anonymized Medidata Clinical Cloud data) from closed clinical trials conducted between 2010-2023 were utilized to train a predictive model. The model leveraged the procedure and schedule-of-activity data to capture the frequency of protocol-specified procedures as input features, and enrollment and dropout rates as outcome features. Model performance was assessed utilizing cross-validation. To assess the relative impact of individual procedures or groups of procedures on a protocol, a sensitivity analysis was performed by increasing each procedure’s frequency by 1 unit while holding all other variables constant and measuring the resulting percent change in predicted enrollment and dropout rates. Results: Analysis of selected protocols demonstrated that inclusion of Positron Emission Tomography (PET) Imaging was associated with a 10-23% reduction in enrollment rates. Genetic testing and analysis procedures were associated with enrollment rate decreases of up to 5%. In contrast, bone density imaging, abdominal and pelvic imaging, and increased tumor and biopsy procedures were associated with enrollment increase of up to 6%. With respect to retention, increased cardiac rhythm monitoring was associated with up to 10% reduction in dropout rates, while blood cell analysis, collagen biomarker testing, and quality of life questionnaires were each associated with dropout reductions of up to 8%, 8% and 5% respectively. Conclusions: Procedure requirements and frequency had a measurable impact on clinical trial enrollment and retention. Some procedures, such as PET scans, were associated with reduced enrollment; whereas other clinically relevant imaging procedures (i.e., DEXA bone density imaging, CT/MRI abdominal and pelvic imaging) and other tumor-related procedures, monitoring, and assessments were associated with improved enrollment and/or reduced dropout. These findings highlight opportunities to optimize protocol design by prioritizing procedures that improve the likelihood of recruitment and retention.